Uncertainty engine: calibration roll cron (#101)#112
Merged
ssilvius merged 2 commits intofeat/uncertainty-orphan-cronfrom Apr 28, 2026
Merged
Uncertainty engine: calibration roll cron (#101)#112ssilvius merged 2 commits intofeat/uncertainty-orphan-cronfrom
ssilvius merged 2 commits intofeat/uncertainty-orphan-cronfrom
Conversation
Closes #101. Phase 1 of the uncertainty engine spike. Pure bucketing in api/lib/uncertainty/calibration.ts: - bucketCohort splits witnessed predictions across 10 buckets of 0.1 width on claimed_confidence. Per non-empty bucket: mean claimed, mean correctness, count, and Brier as mean squared error. Per-cohort orphan_count is attached to every emitted snapshot row so the dashboard surfaces the orphan share alongside reliability. - rollCalibration walks every active cohort (predictions in witnessed | orphaned), atomically replaces prior snapshot rows for the cohort, and returns counts of cohorts and rows touched. Exposed at POST /api/uncertainty/internal/calibration-roll, sharing the CRON_SECRET bearer guard with the orphan sweep. wrangler.jsonc adds a 03:15 UTC daily trigger alongside the hourly orphan sweep; both are invoked by an external scheduler against the internal endpoints since the Astro adapter does not expose a scheduled handler. Tests cover empty cohorts, single-bucket and multi-bucket spreads, mean / Brier math, null-correctness rows, the confidence=1.0 edge, and the orphan-attachment invariant (10 calibration tests). Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Closes #101. Stacks on #111.
Pure bucketing in api/lib/uncertainty/calibration.ts:
Exposed at POST /api/uncertainty/internal/calibration-roll, sharing the CRON_SECRET bearer guard with the orphan sweep. wrangler.jsonc adds a 03:15 UTC daily trigger alongside the hourly orphan sweep; both are invoked by an external scheduler against the internal endpoints since the Astro adapter does not expose a scheduled handler.
Tests cover empty cohorts, single-bucket and multi-bucket spreads, mean / Brier math, null-correctness rows, the confidence=1.0 edge, and the orphan-attachment invariant (10 calibration tests).